• Chinese Journal of Lasers
  • Vol. 50, Issue 13, 1304005 (2023)
Xing Hu1、3、4, Shangbin Yang1、3、4、*, Kaifan Ji2、4, Jiaben Lin1、3、4, Yuanyong Deng1、3、4, Xianyong Bai1、3、4, Xiaoming Zhu1、3, Yang Bai1、3, and Quan Wang1、3、4
Author Affiliations
  • 1National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, China
  • 2Yunnan Observatories, Chinese Academy of Sciences, Kunming 650217, Yunnan, China
  • 3Key Laboratory of Solar Activity, Chinese Academy of Sciences, Beijing 100101, China
  • 4University of Chinese Academy of Sciences, Beijing 100049, China
  • show less
    DOI: 10.3788/CJL221432 Cite this Article Set citation alerts
    Xing Hu, Shangbin Yang, Kaifan Ji, Jiaben Lin, Yuanyong Deng, Xianyong Bai, Xiaoming Zhu, Yang Bai, Quan Wang. Calibration of Observing Wavelength Points of Birefringent Narrow Band Filter-Type Magnetograph Based on Neural Network[J]. Chinese Journal of Lasers, 2023, 50(13): 1304005 Copy Citation Text show less
    Spectral line profile obtained by fitting of data
    Fig. 1. Spectral line profile obtained by fitting of data
    Effect of Doppler velocity of solar autorotation on images at different points of spectral line
    Fig. 2. Effect of Doppler velocity of solar autorotation on images at different points of spectral line
    Comparison before and after data processing (wavelengths in nm). (a) Original images; (b) grayscale change of original images; (c) images after edge dimming is removed; (d) grayscale change after edge dimming is removed
    Fig. 3. Comparison before and after data processing (wavelengths in nm). (a) Original images; (b) grayscale change of original images; (c) images after edge dimming is removed; (d) grayscale change after edge dimming is removed
    Removing information outside solar circle. (a) Original image;(b) image after polar coordinate transformation;(c) solar part
    Fig. 4. Removing information outside solar circle. (a) Original image;(b) image after polar coordinate transformation;(c) solar part
    PCA decomposition contribution rate
    Fig. 5. PCA decomposition contribution rate
    Image comparison before and after 20th-order reconstruction
    Fig. 6. Image comparison before and after 20th-order reconstruction
    Image pre-processing process
    Fig. 7. Image pre-processing process
    MLP regression network
    Fig. 8. MLP regression network
    General process of calibration of observing wavelength points using BP neural network
    Fig. 9. General process of calibration of observing wavelength points using BP neural network
    Correlation diagram of training set and test set for method validation experiment. (a) Training set; (b) test set
    Fig. 10. Correlation diagram of training set and test set for method validation experiment. (a) Training set; (b) test set
    Results of group 1 of grouping test experiment. (a) Correlation diagram of training set;(b) correlation diagram of test set;(c) residual of each data prediction result;(d) variation of residual standard deviation of each day with time
    Fig. 11. Results of group 1 of grouping test experiment. (a) Correlation diagram of training set;(b) correlation diagram of test set;(c) residual of each data prediction result;(d) variation of residual standard deviation of each day with time
    Variation of residual standard deviation of different groups of grouping test experiment. (a) Group 2; (b) group 3; (c) group 4; (d) group 5
    Fig. 12. Variation of residual standard deviation of different groups of grouping test experiment. (a) Group 2; (b) group 3; (c) group 4; (d) group 5
    Test results in range of [-0.006 nm, 0.006 nm]
    Fig. 13. Test results in range of [-0.006 nm, 0.006 nm]
    HardwareSpecification
    CPUIntel Core i9-9900KF @ 3.60 GHz 8-core
    RAM64 GB
    GPUNVIDIA GeForce RTX 2070(8 GB)
    MainboardASUS TUF Z390-PLUS GAMING
    Table 1. Experimental hardware environment
    Traditional methodMachine learning method
    Image acquisition timeSpectral line fitting timeTotal timeImage acquisition timeTime for obtaining corresponding wavelengthTotal time
    14-19 minAbout 60 s15-20 min5 s2 s7 s
    Table 2. Time comparison between traditional calibration method and machine learning calibration method
    Xing Hu, Shangbin Yang, Kaifan Ji, Jiaben Lin, Yuanyong Deng, Xianyong Bai, Xiaoming Zhu, Yang Bai, Quan Wang. Calibration of Observing Wavelength Points of Birefringent Narrow Band Filter-Type Magnetograph Based on Neural Network[J]. Chinese Journal of Lasers, 2023, 50(13): 1304005
    Download Citation